| { | |
| "arc": { | |
| "num_heads": 12, | |
| "attention_dim": 768, | |
| "vocab_size": 43012, | |
| "num_blocks": 12, | |
| "ff_dim": 2304 | |
| }, | |
| "attn": { | |
| "attn_chunks": 2, | |
| "gqa_repeats": 3, | |
| "theeta": 10000 | |
| }, | |
| "moe": { | |
| "num_experts": 1, | |
| "top_k": 1, | |
| "capacity": 1.0 | |
| }, | |
| "dropout_rate": 0.05 | |
| } |